CN108229305B - Method and device for determining outer frame of target object and electronic equipment - Google Patents

Method and device for determining outer frame of target object and electronic equipment Download PDF

Info

Publication number
CN108229305B
CN108229305B CN201711165979.8A CN201711165979A CN108229305B CN 108229305 B CN108229305 B CN 108229305B CN 201711165979 A CN201711165979 A CN 201711165979A CN 108229305 B CN108229305 B CN 108229305B
Authority
CN
China
Prior art keywords
attribute information
key point
key points
target object
information
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201711165979.8A
Other languages
Chinese (zh)
Other versions
CN108229305A (en
Inventor
李步宇
李全全
闫俊杰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Sensetime Technology Development Co Ltd
Original Assignee
Beijing Sensetime Technology Development Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Sensetime Technology Development Co Ltd filed Critical Beijing Sensetime Technology Development Co Ltd
Priority to CN201711165979.8A priority Critical patent/CN108229305B/en
Publication of CN108229305A publication Critical patent/CN108229305A/en
Priority to SG11201913529UA priority patent/SG11201913529UA/en
Priority to PCT/CN2018/111464 priority patent/WO2019100886A1/en
Priority to JP2019572712A priority patent/JP6872044B2/en
Priority to US16/731,858 priority patent/US11348275B2/en
Application granted granted Critical
Publication of CN108229305B publication Critical patent/CN108229305B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/048—Activation functions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
    • G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/103—Static body considered as a whole, e.g. static pedestrian or occupant recognition
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20076—Probabilistic image processing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30196—Human being; Person
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2210/00—Indexing scheme for image generation or computer graphics
    • G06T2210/12—Bounding box

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Multimedia (AREA)
  • Computing Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Molecular Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Medical Informatics (AREA)
  • Human Computer Interaction (AREA)
  • Image Analysis (AREA)

Abstract

The application discloses a method, a device, a computer-readable storage medium and an electronic device for determining a circumscribed frame of a target object, wherein the method comprises the following steps: acquiring attribute information of each key point in a plurality of key points of a target object; and determining the position of an outer frame of the target object according to the attribute information of each key point in the plurality of key points of the target object and a preset neural network. According to the method and the device, the efficiency and the accuracy of determining the outer frame of the target object can be improved.

Description

Method and device for determining outer frame of target object and electronic equipment
Technical Field
The present application relates to computer vision technology, and in particular, to a method, an apparatus, and an electronic device for determining a bounding box of a target object.
Background
In the field of computer vision such as image recognition, it is often necessary to quickly and accurately determine an external frame of a human body.
At present, a circumscribed frame of a human body is usually determined by using fast-RCNN (accelerated Convolutional Neural Networks), that is, RPN (Region-candidate Network) is used to obtain a plurality of candidate regions, and then, RCNN is used to score and correct each candidate Region, so as to determine the circumscribed frame of the human body.
Disclosure of Invention
The embodiment of the application provides a method for determining a circumscribed frame of a target object, a device for determining the circumscribed frame of the target object, a computer-readable storage medium and an electronic device.
According to an aspect of the present application, there is provided a method for determining a bounding box of a target object, comprising: acquiring attribute information of each key point in a plurality of key points of a target object; and determining the position of an outer frame of the target object according to the attribute information of each key point in the plurality of key points of the target object and a preset neural network.
In an embodiment of the present application, the target object includes: a human body.
In another embodiment of the present application, the attribute information of the key point includes: coordinate information and presence discrimination values.
In another embodiment of the present application, the determining, according to the attribute information of each of the plurality of key points of the target object and a preset neural network, a position of an outline box of the target object includes: determining at least one effective key point from the plurality of key points according to the attribute information of each key point in the plurality of key points; processing the attribute information of the plurality of key points according to the attribute information of each effective key point in the at least one effective key point to obtain the processed attribute information of the plurality of key points; and inputting the processed attribute information of the plurality of key points into the preset neural network for processing to obtain the position of the outer frame of the target object.
In another embodiment of the present application, the processed attribute information of the plurality of key points includes: the processed attribute information of each effective key point in the at least one effective key point and the attribute information of other key points except the at least one effective key point in the plurality of key points.
In another embodiment of the present application, the processing the attribute information of the plurality of key points according to the attribute information of each of the at least one valid key point to obtain processed attribute information of the plurality of key points includes: determining a reference coordinate according to coordinate information included in the attribute information of each effective key point in the at least one effective key point; and determining coordinate information in the processed attribute information of each effective key point according to the reference coordinate and the coordinate information in the attribute information of each effective key point in the at least one effective key point.
In another embodiment of the present application, the determining the reference coordinate according to the coordinate information included in the attribute information of each valid keypoint of the at least one valid keypoint includes: carrying out average processing on coordinates corresponding to the coordinate information of each effective key point in the at least one effective key point to obtain the reference coordinates; and/or determining coordinate information in the processed attribute information of each effective key point according to the reference coordinate and the coordinate information in the attribute information of each effective key point in the at least one effective key point, wherein the determining comprises the following steps: and determining the processed coordinate information corresponding to the coordinate information of each effective key point in the at least one effective key point by taking the reference coordinate as an origin.
In another embodiment of the present application, the inputting the processed attribute information of the plurality of key points into the preset neural network for processing to obtain the position of the outer frame of the target object includes: inputting the processed attribute information of the plurality of key points into the preset neural network for processing to obtain output position information; and determining the position of an outer frame of the target object according to the reference coordinate and the output position information.
In yet another embodiment of the present application, the method further comprises: obtaining a sample set comprising a plurality of sample data, wherein the sample data comprises: attribute information of a plurality of key points of a sample object, wherein the sample data is marked with an outer frame position of the sample object;
and training the neural network according to the attribute information of a plurality of key points of the sample object in each sample data and the position of the external frame of the sample object.
In yet another embodiment of the present application, the neural network is trained based on a stochastic gradient descent algorithm.
In another embodiment of the present application, the position of the outer frame of the target object includes: and coordinate information of two vertexes in the diagonal direction of the circumscribed frame of the target object.
In yet another embodiment of the present application, the neural network includes: at least two fully-connected layers.
In yet another embodiment of the present application, the neural network includes: a three-layer fully-connected layer, wherein an activation function of at least one of a first layer fully-connected layer and a second layer fully-connected layer of the three-layer fully-connected layer comprises: the linear element ReLu activation function is modified.
In yet another embodiment of the present application, the first fully-connected layer includes 320 neurons, the second neural network includes 320 neurons, and the last fully-connected layer of the three fully-connected layers includes 4 neurons.
According to another aspect of the present application, there is provided an apparatus for determining a bounding box of a target object, comprising: the acquisition module is used for acquiring the attribute information of each key point in a plurality of key points of the target object; and the determining module is used for determining the position of the outer frame of the target object according to the attribute information of each key point in the plurality of key points of the target object acquired by the acquiring module and a preset neural network.
In particular, the apparatus may be adapted to perform any of the embodiments of the method described above, and accordingly, the apparatus may comprise means and/or modules for performing any of the embodiments of the method described above.
According to still another aspect of the present application, there is provided an electronic device including: a processor and a computer readable storage medium for storing instructions, execution of which by the processor causes the electronic device to perform any of the embodiments of the method described above.
According to yet another aspect of the present application, there is provided a computer readable storage medium having stored thereon instructions that, when executed by a processor, perform any of the above-described method embodiments.
According to yet another aspect of the application, there is provided a computer program product comprising at least one instruction which, when executed by a processor, performs any of the embodiments of the method described above.
Based on the method and the device for determining the circumscribed frame of the target object and the electronic device, provided by the above embodiments of the present application, the position of the circumscribed frame of the target object is determined by using the attribute information of each of the plurality of key points of the target object and the neural network, which is beneficial to improving the efficiency and the accuracy of determining the circumscribed frame of the target object.
The technical solution of the present application is further described in detail by the accompanying drawings and embodiments.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the application and together with the description, serve to explain the principles of the application.
FIG. 1 is a flow chart of one embodiment of a method in the present application;
FIG. 2 is a flow diagram of one embodiment of a method of training a neural network in the present application;
FIG. 3 is a schematic diagram of the detailed structure of one embodiment of the apparatus of the present application;
FIG. 4 is a schematic diagram of an embodiment of an electronic device;
FIG. 5 is a schematic diagram of one embodiment of a computer-readable storage medium in the present application.
Detailed Description
Various exemplary embodiments of the present application will now be described in detail with reference to the accompanying drawings. It should be noted that: the relative arrangement of the components and steps, the numerical expressions, and numerical values set forth in these embodiments do not limit the scope of the present application unless specifically stated otherwise.
Meanwhile, it should be understood that the sizes of the respective portions shown in the drawings are not drawn in an actual proportional relationship for the convenience of description.
The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the application, its application, or uses.
Techniques, methods, and apparatus known to those of ordinary skill in the relevant art may not be discussed in detail but are intended to be part of the specification where appropriate.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, further discussion thereof is not required in subsequent figures.
The embodiments of the application are applicable to computer systems/servers operable with numerous other general purpose or special purpose computing system environments or configurations. Examples of well known computing systems, environments, and/or configurations that may be suitable for use with the computer system/server include, but are not limited to: personal computer systems, server computer systems, thin clients, thick clients, hand-held or laptop devices, microprocessor-based systems, set top boxes, programmable consumer electronics, network pcs, minicomputer systems, mainframe computer systems, distributed cloud computing environments that include any of the above systems, and the like.
The computer system/server may be described in the general context of computer system-executable instructions, such as program modules, being executed by a computer system. Generally, program modules may include routines, programs, objects, components, logic, and data structures, etc. that perform particular tasks or implement particular abstract data types. The computer system/server may be practiced in distributed cloud computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed cloud computing environment, program modules may be located in both local and remote computer system storage media including memory storage devices.
Various non-limiting embodiments of the present application are described in detail below.
Fig. 1 is a schematic flow chart of a method for determining a bounding box of a target object according to the present application. As shown in fig. 1, the method for determining a bounding box of a target object of the present application mainly includes: s100 and S110. The steps in fig. 1 will be described below.
S100, acquiring attribute information of each key point in a plurality of key points of the target object.
In an optional example, a target object in the embodiment of the present application may also be referred to as a detection object or an outer frame detection object, which is not limited in this application example. Optionally, the target object may be a human body, or may also be a human face or a certain specific object, and the embodiment of the present application does not limit the specific representation form of the target object. The bounding box in the embodiments of the present application generally refers to a polygon (generally a rectangle) that can represent the region where the target object is located, and the bounding box can not only accurately cover all parts of the target object, but also has an area as small as possible.
In an alternative example, the attribute information of the key point in the embodiment of the present application may include various information of the key point. As an example, the attribute information of the key points may be used to describe whether each key point of the target object is visible in the image and the specific position of each key point visible in the image; the embodiments of the present application may refer to keypoints that are visible in the image (i.e., keypoints located in the image) as valid keypoints, and refer to keypoints that are not visible in the image (i.e., keypoints not located in the image) as invalid keypoints. The invisible key points in the image may be the blocked key points or the key points located outside the image, which is not limited in the embodiment of the present application.
In a specific example, the attribute information of the key point may include: coordinate information of the key point and an existence discrimination value of the key point, where the coordinate information of the key point may be used to represent a position of the key point in the image, for example, the coordinate information of the key point may specifically be a two-dimensional coordinate of the key point, but the embodiment of the present application is not limited thereto; the keypoint presence decision value may be used to indicate whether the keypoint is visible in the image. For example, if the existence criterion value of a key point is 1, it indicates that the key point is visible, and if the existence criterion value of a key point is 0, it indicates that the key point is invisible, but the existence criterion value in the embodiment of the present application may be implemented in other ways. Optionally, the attribute information may further include other information, and the embodiment of the present application is not limited thereto.
As an example, the attribute information of the keypoints acquired in the embodiment of the present application may be a 3 × N-dimensional vector, where N represents the number of multiple keypoints that are target objects in advance. In the embodiment of the present application, the attribute information of a keypoint may be represented by an array (x, y, v), where x and y are two-dimensional coordinates of the keypoint in the image, and v is an existence criterion value of the keypoint, when a value of v is a first criterion value, the keypoint is a visible keypoint in the image, and when a value of v is a second criterion value, the keypoint is an invisible keypoint in the image. For example, for one keypoint of the target object, if the keypoint is a valid keypoint, the attribute information of the keypoint may be represented as an array (x, y, 1), and if the keypoint is an invalid keypoint (occluded or located outside the image), the attribute information of the keypoint may be represented as an array (0, 0, 0). By representing the attribute information of the key points in such a way, the actual situation of all the key points of the target object in the image can be conveniently known.
In an alternative example, in the case that the target object is a human body, the key points of the human body of the embodiment of the present application may generally include: the 14 key points can be used for completely describing the posture and the shape of the human body. At this time, the attribute information of the plurality of key points may include attribute information of some or all of the 14 key points. As an example, the attribute information of the plurality of key points acquired in the embodiment of the present application may include: the coordinate information of the vertex and the existence determination value of the vertex, the coordinate information of the neck and the existence determination value of the neck, the coordinate information of the left shoulder and the existence determination value of the left shoulder, the coordinate information of the right shoulder and the existence determination value of the right shoulder, the coordinate information of the left elbow and the existence determination value of the left elbow, the coordinate information of the right elbow and the existence determination value of the right elbow, the coordinate information of the left wrist and the existence determination value of the left wrist, the coordinate information of the left hip and the existence determination value of the left hip, the coordinate information of the right hip and the existence determination value of the right hip, the coordinate information of the left knee and the existence determination value of the right knee, the coordinate information of the left ankle and the existence determination value of the left ankle, and the coordinate information of the right ankle and the existence determination value of the right ankle. By using the attribute information of the 14 key points, the human body profile in the image can be described. In the case where the target object is something else, the key points thereof often change accordingly, and the embodiment of the present application does not limit the concrete expression form of the key points of the target object.
In an alternative example, the embodiment of the present application may be applied to an application scenario in which attribute information of a plurality of key points of a target object has been obtained, that is, in an application scenario in which attribute information of a plurality of key points of a target object has been obtained from an image or in another manner, the embodiment of the present application may obtain the attribute information of the key points of the target object by means of information reading or the like, but the embodiment of the present application is not limited thereto. In such an application scenario, according to the attribute information of the plurality of key points of the target object, the embodiment of the present application may obtain the position of the outer frame of the target object by using a neural network trained in advance.
S110, determining the position of an external frame of the target object according to the attribute information of each key point in the plurality of key points of the target object and a preset neural network.
The bounding box position of the target object may be used to determine the bounding box of the target object. Optionally, the bounding box location may include location information for one or more vertices of the bounding box. In an alternative example, if the bounding box is a quadrilateral, such as a rectangle, the bounding box position may include position information of two opposite vertices of the bounding box, for example, two-dimensional coordinates of each of the two opposite vertices, but the embodiment of the present application does not limit the specific implementation of the bounding box position of the target object.
In an alternative example, the neural network in embodiments of the present application may be a dedicated neural network. The neural network may be trained using a large amount of sample data, where the sample data may include attribute information of a plurality of key points of the sample object and a position of the circumscribed box, that is, the sample data may be labeled with the position of the circumscribed box of the sample object. A specific example of the training process can be seen in the following description with respect to fig. 2, and will not be described in detail here.
Optionally, the neural network in the embodiment of the present application may include: at least two fully-connected layers. Fully-connected networks may have faster computational speed and processing efficiency than convolutional neural networks.
In an alternative example, the neural network in the embodiment of the present application includes: two full connection layers, and the activation function of the first full connection layer may be a reli (Rectified Linear Unit) activation function.
In an alternative example, the neural network in the embodiment of the present application includes: and the activation function of the first full connection layer can be a ReLu activation function, and the activation function of the second full connection layer can also be a ReLu activation function.
In the embodiment of the present application, the number of fully-connected layers included in the neural network and the number of neurons included in each fully-connected layer may be set according to actual conditions. Under the condition that the number of layers of the neural network and the number of the neurons are enough, the neural network has stronger function expression capacity, so that the position of the outer frame obtained based on the neural network is more accurate. In an alternative example, for a neural network formed by three fully-connected layers, the number of neurons in a first fully-connected layer of the neural network may be 320, the number of neurons in a second fully-connected layer of the neural network may be 320, and the number of neurons in a third fully-connected layer may be set to 4 in a case where the circumscribed frame of the target object is a quadrangle (e.g., a rectangle) and the position of the circumscribed frame of the target object is represented by two-dimensional coordinate information of two vertices in the diagonal direction of the circumscribed frame.
The inventor verifies through a plurality of experiments that the three-layer full-connection layer is provided, the ReLu activation function is used for the activation functions of the first layer and the second layer full-connection layer, the first layer and the second layer full-connection layer are respectively provided with 320 neurons, the third layer full-connection layer is provided with a neural network with 4 neurons, the operation speed of the neural network can meet the actual requirement, and the accuracy of determining the position of the external frame can also meet the actual requirement.
In the embodiment of the present application, the attribute information of the plurality of key points may be directly input to the neural network, or may be input to the neural network after processing the attribute information of the plurality of key points. That is, the input information of the neural network may be determined according to the attribute information of the plurality of key points, where the input information may be the attribute information of the plurality of key points itself or obtained by processing the attribute information of the plurality of key points. The neural network can process the input information to obtain an output result, wherein the position of the outer frame of the target object can be obtained according to the output result of the neural network. Specifically, the output result of the neural network may include position information of the bounding box of the target object, for example, coordinate information of one or more vertices of the bounding box of the target object, and as an example, if the bounding box is a rectangle, the output result may include coordinate information of two opposite vertices of the bounding box; alternatively, the position of the outer frame of the target object may be obtained by processing an output result of the neural network, which is not limited in this embodiment of the present application.
In an optional example, the method and the device can select the effective key points according to the attribute information of each key point in a plurality of key points of the target object. For example, if the attribute information of the keypoint includes a presence discrimination value, the keypoint whose presence discrimination value indicates presence may be determined as a valid keypoint, for example, if the presence discrimination value of the keypoint is 1, the keypoint may be determined as a valid keypoint, but the embodiment of the present application is not limited thereto.
Optionally, if at least one valid key point can be selected from the plurality of key points, the attribute information of some or all of the plurality of key points may be processed according to the attribute information of each valid key point in the at least one valid key point to obtain the processed attribute information of the plurality of key points, and the processed attribute information of the plurality of key points is used as the input information. Specifically, the processed attribute information of the plurality of key points may include the processed attribute information of each of the plurality of key points, or include the processed attribute information of a part of the plurality of key points and the original attribute information of another part of the plurality of key points. As an example, the processed attribute information of the plurality of key points may include the processed attribute information of each valid key point in the at least one valid key point and the original attribute information of other key points in the plurality of key points except the at least one valid key point, that is, the attribute information of each valid key point in the at least one valid key point may be processed without processing the attribute information of other key points, but the embodiment of the present application is not limited thereto.
In the embodiment of the present application, the attribute information of the at least one valid keypoint may be processed in various ways. As an example, the reference coordinate may be determined according to coordinate information included in the attribute information of each of the at least one valid key point, and the coordinate information in the processed attribute information of the valid key point may be determined according to the coordinate information in the attribute information of the valid key point and the reference coordinate. The reference coordinates may be obtained by processing coordinate information of the at least one valid keypoint. For example, the reference coordinate may be obtained by averaging the coordinates of the at least one valid keypoint, but the embodiment of the present application does not limit the specific implementation of the reference coordinate.
In an optional example, the attribute information of the keypoint acquired in S100 may be subjected to zero-averaging processing, and information obtained after the zero-averaging processing is provided to the neural network as a part of the input information. For example, a coordinate mean (m) may be calculated from coordinate information in the attribute information of the effective keypointsx，my) (ii) a Then, for each valid key point of all key points, the coordinate information (x) of the key point is calculated respectivelyi，yi) Difference from the above coordinate mean, i.e. (x)i-mx，yi-my) And using the calculated difference as the valueCoordinate information of the effective key points; finally, the coordinate information of all the key points of the target object and the presence discrimination values of all the key points may be supplied as input information to the neural network.
It should be noted that, if the embodiment of the present application does not perform zero-mean processing on the sample data in the process of training the neural network, in S110, it is also not necessary to perform zero-mean processing on the two-dimensional coordinates of the valid keypoints of the target object.
Alternatively, in the case where the input information provided to the neural network is input information subjected to zero-mean processing, the embodiment of the present application may sum the coordinate information output by the neural network and the calculated coordinate mean to obtain the final coordinates of a plurality of vertices (e.g., two vertices on a diagonal line of a rectangular bounding box) of the bounding box as the target object. For example, the output position information of the neural network is (bx)1，by1) And (bx)2，by2) Then the coordinates of the two vertices on the bounding box diagonal of the target object may be (bx)1+mx，by1+my) And (bx)2+mx，by2+my)。
Fig. 2 is a schematic flow chart of a method for training a neural network according to an embodiment of the present disclosure. Here, it is assumed that the number of the plurality of key points is N, and the attribute information of each key point may be a 3-dimensional vector: (x, y, v), the circumscribing box is rectangular. Further, assume that the input to the neural network comprises a 3 × N matrix and the output comprises a 2 × 2 matrix, in particular two-dimensional coordinates of two vertices on a diagonal of the bounding box.
As shown in fig. 2, a method for training a neural network according to an embodiment of the present disclosure may include: s200, S210, S220, S230, S240, and S250. The steps in fig. 2 will be described below.
S200, obtaining a piece of sample data from the sample set.
In an alternative example, the sample set in the embodiment of the present application is generally non-empty and generally includes a large amount of sample data, for example, the sample set may be embodied in the presently disclosed MS COCO database or the like. Each sample data in the sample set may include: the method includes the steps of obtaining attribute information of a plurality of key points of a sample object, and labeling each sample data with a position of a bounding box of the sample object, where the attribute information of the key points may include coordinate information of the key points and presence decision values of the key points, but the embodiment of the present application is not limited thereto. The sample object corresponding to the sample data is generally of the same type as the target object, and for example, when the target object is a human body, the sample object is also a human body. According to the embodiment of the application, one piece of sample data can be selected from the sample set according to the arrangement sequence of the sample data, and one piece of sample data can also be selected randomly from the sample set.
S210, calculating a coordinate mean value according to the coordinate information of all valid key points of the sample data, for example, calculating a coordinate mean value (m) for the coordinate information attributes of all key points with v being 1 in the sample datax，my)。
S220, calculating a difference between the coordinate information of each valid keypoint in the sample data and the coordinate mean, and using the calculated difference as the coordinate information of the corresponding valid keypoint, that is, the coordinate information (x) of the valid keypoint in the sample datai，yi) Calculating (x)i-mx，yi-my)。
And S230, providing the attribute information of all key points of the sample data as input to a neural network.
In one alternative example, the output of the neural network is the two-dimensional coordinates (bx) of two vertices on a diagonal of a rectangle1，by1) And (bx)2，by2) In this case, the coordinates of the bounding box may be determined as the difference between the output coordinate information and the coordinate mean (i.e. given supervision), and may be specifically expressed as (bx)1-mx，by1-my) And (bx)2-mx，by2-my)。
Optionally, the embodiment of the present application may use a stochastic gradient descent algorithm for calculation to implement training.
Optionally, whether to adjust the parameters of the neural network may be determined by comparing the result calculated by the neural network with the position of the outer frame labeled by the sample data. If the difference between the result obtained by the calculation of the neural network and the position of the outer frame marked by the sample data is lower than a certain range, the training process can be terminated or new sample data can be continuously selected from the sample set. Otherwise, the parameters of the neural network can be adjusted, and the calculation is continued by using the adjusted neural network.
And S240, judging whether to continue to obtain a new sample data from the sample set, if so, returning to S200, otherwise, returning to S250.
In an optional example, the embodiment of the present application may determine whether to continue to obtain a new piece of sample data from the sample set by determining whether all sample data in the sample set is used for training, whether a result output by the neural network meets a predetermined accuracy requirement, or whether the number of read samples reaches a predetermined number.
And S250, finishing the training process.
In an alternative example, by detecting, in a case that it is determined that a result output by the neural network meets a predetermined accuracy requirement, the training of the neural network is successful, and if all sample data in the sample set have been used for training or the number of read samples has reached a predetermined number, however, by detecting that the result output by the neural network does not meet the predetermined accuracy requirement yet, although the training process is ended, the training of the neural network is not successful, the neural network may be trained again. The detection may specifically be: selecting a plurality of untrained sample data from the sample set, providing the sample data to the neural network according to the method shown in fig. 1 based on the sample data, determining the error between each circumscribed frame position obtained based on the neural network and the manually marked circumscribed frame position in the corresponding sample data, and successfully training the neural network when the accuracy meets the preset accuracy requirement according to each error. In addition, in the embodiment of the present application, in the process of training the neural network, the L2 loss function may be used for training supervision, but the embodiment of the present application is not limited thereto.
The method and the device train the neural network by utilizing the attribute information of the key points of the sample object and the position of the external frame, so that the trained neural network can directly determine the position of the external frame of the target object based on the attribute information of the key points of the target object; in some practical applications, the attribute information of the key points of the target object is successfully obtained, so that the external frame of the target object can be quickly obtained by fully utilizing the obtained attribute information of the key points of the target object under the condition of not needing an image; because the neural network in the embodiment of the application is obtained by using the attribute information of the key points of the sample object and the position training of the external frame, under the condition that the number of the key points of the sample object is large and the number of the set neurons is large, the number of the parameters required to be learned by the neural network is also large, which is beneficial to enabling the neural network to more accurately determine the external frame of the target object.
Fig. 3 is a schematic structural diagram of an embodiment of an apparatus for determining a bounding box of a target object according to the present application. The apparatus of this embodiment may be used to implement the method embodiments described above in this application.
As shown in fig. 3, the apparatus of this embodiment mainly includes: the obtaining module 300 and the determining module 310 may further include: a training module 320.
The obtaining module 300 is mainly used for obtaining attribute information of each key point in a plurality of key points of a target object. Specific contents of the target object, the key points, the valid key points, and the attribute information of the key points can be referred to the related description of S100 in the above method embodiment, and are not described in detail here.
In an alternative example, the apparatus of the embodiment of the present application is generally applicable to an application scenario in which the attribute information of the key points of the target object has been successfully obtained, that is, in an application scenario in which the attribute information of the key points of the target object has been obtained from an image, the obtaining module 300 may directly obtain the attribute information of the key points of the target object that already exists by reading or the like.
The determining module 310 is mainly configured to determine the position of the circumscribed frame of the target object according to the attribute information of each key point of the plurality of key points of the target object acquired by the acquiring module 300 and a preset neural network. The specific representation of the neural network (e.g., the number of layers, the number of neurons, the activation function, etc.) in the present application can be referred to the related description of the above method embodiments, and will not be described in detail herein.
In one optional example, the determining module 310 may include: a first sub-module, a second sub-module, and a third sub-module. The first sub-module is mainly configured to determine at least one valid keypoint from the multiple keypoints according to the attribute information of each keypoint of the multiple keypoints acquired by the acquisition module 300. The second submodule is mainly used for processing the attribute information of the plurality of key points according to the attribute information of each effective key point in the at least one effective key point determined by the first submodule to obtain the processed attribute information of the plurality of key points. The third sub-module is mainly used for inputting the processed attribute information of the plurality of key points obtained by the second sub-module into a preset neural network for processing to obtain the position of the outer frame of the target object. The processed attribute information of the plurality of key points in the present application may include: the processed attribute information of each of the at least one valid keypoint and the attribute information of other keypoints of the plurality of keypoints other than the at least one valid keypoint. The second sub-module may specifically include: a first unit and a second unit. The first unit is mainly used for determining a reference coordinate according to coordinate information included in the attribute information of each effective key point in at least one effective key point determined by the first submodule; for example, the first unit performs an averaging process on coordinates corresponding to the coordinate information of each effective key point in the at least one effective key point to obtain a reference coordinate. The second unit is mainly used for determining the coordinate information in the processed attribute information of each effective key point according to the reference coordinate obtained by the first unit and the coordinate information in the attribute information of each effective key point in at least one effective key point; for example, the second unit determines the processed coordinate information corresponding to the coordinate information of each effective key point in the at least one effective key point, using the reference coordinate obtained by the first unit as an origin. At this time, the third sub-module may be specifically configured to input the attribute information of the plurality of key points processed by the second unit to the neural network for processing, to obtain output position information, and determine the position of the circumscribed frame of the target object according to the reference coordinate and the output position information.
In an optional example, in a case where zero-mean processing needs to be performed on two-dimensional coordinates of key points of the target object, the first unit is configured to calculate a two-dimensional coordinate mean value according to coordinate information of all valid key points of the target object; the second unit is used for respectively calculating the difference value between the coordinate information of the key point and the mean value of the two-dimensional coordinates aiming at all effective key points of the target object, and taking the difference value as the coordinate information of the effective key points; and the third submodule is used for providing the coordinate information of all key points of the target object and the existence discrimination values of all key points as input information to the neural network.
In the case that the determining module 310 performs zero-mean processing on the two-dimensional coordinates of the key points of the target object, the determining module 310 may use the sum of the circumscribed frame coordinate information output by the neural network and the coordinate mean as the circumscribed frame two-dimensional coordinate information of the target object.
The training module 320 is mainly used for training the neural network.
In an optional example, the training module 320 obtains a plurality of pieces of sample data from the sample set, for each piece of sample data, the training module 320 calculates a coordinate mean value according to coordinate information of all effective key points of the piece of sample data, respectively calculates a difference value between the coordinate information of each effective key point in the piece of sample data and the coordinate mean value, and provides the calculated difference value as coordinate information of a corresponding effective key point, and the training module 320 provides attribute information of all key points of the piece of sample data as input to the neural network. One specific example of the operation performed by the training module 320 to train the neural network can be found in the description of the method embodiment, and will not be repeated here.
The embodiment of the application also provides an electronic device, which can be a mobile terminal, a Personal Computer (PC), a tablet computer, a server and the like. Referring now to fig. 4, there is shown a schematic diagram of an electronic device 400 suitable for use in implementing a terminal device or server of an embodiment of the present application: as shown in fig. 4, the computer system 400 includes one or more processors, communication sections, and the like, for example: one or more Central Processing Units (CPUs) 401, and/or one or more image processors (GPUs) 413, etc., which may perform various appropriate actions and processes according to executable instructions stored in a Read Only Memory (ROM)402 or loaded from a storage section 408 into a Random Access Memory (RAM) 403. The communication section 412 may include, but is not limited to, a network card, which may include, but is not limited to, an ib (infiniband) network card. The processor may communicate with the read only memory 402 and/or the random access memory 430 to execute the executable instructions, connect with the communication part 412 through the bus 404, and communicate with other target devices through the communication part 412, thereby completing the operations corresponding to any one of the methods provided by the embodiments of the present application.
In addition, in the RAM403, various programs and data necessary for the operation of the apparatus can be stored. The CPU401, ROM402, and RAM403 are connected to each other via a bus 404. The ROM402 is an optional module in case of the RAM 403. The RAM403 stores or writes executable instructions into the ROM402 at runtime, and the executable instructions cause the processor 401 to execute operations corresponding to the above-described communication method. An input/output (I/O) interface 405 is also connected to bus 404. The communication unit 412 may be integrated, or may be provided with a plurality of sub-modules (e.g., a plurality of IB network cards) and connected to the bus link.
The following components are connected to the I/O interface 405: an input section 406 including a keyboard, a mouse, and the like; an output section 407 including a display device such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage section 408 including a hard disk and the like; and a communication section 409 including a network interface card such as a LAN card, a modem, or the like. The communication section 409 performs communication processing via a network such as the internet. A driver 410 is also connected to the I/O interface 405 as needed. A removable medium 411 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 410 as necessary, so that a computer program read out therefrom is mounted into the storage section 408 as necessary.
It should be noted that the architecture shown in fig. 4 is only an optional implementation manner, and in a specific practical process, the number and types of the components in fig. 4 may be selected, deleted, added or replaced according to actual needs; in different functional component settings, separate settings or integrated settings may also be used, for example, the GPU and the CPU may be separately set or the GPU may be integrated on the CPU, the communication part may be separately set or integrated on the CPU or the GPU, and so on. These alternative embodiments are all within the scope of the present disclosure.
In particular, according to an embodiment of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments disclosed herein include a computer program product comprising a computer program tangibly embodied on a computer-readable medium, the computer program comprising program code for performing the method illustrated in the flowcharts, the program code may include instructions corresponding to performing the method steps provided by embodiments of the present application. In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 409, and/or installed from the removable medium 411. The instructions in the computer program, when executed by a Central Processing Unit (CPU)401, perform the above-described functions defined in the method of the present application.
The method and apparatus, device of the present application may be implemented in a number of ways. For example, the methods and apparatuses, devices of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, firmware. The above-described order for the steps of the method is for illustration only, and the steps of the method of the present application are not limited to the order specifically described above unless specifically stated otherwise. Further, in some embodiments, the present application may also be embodied as a program recorded in a recording medium, the program including computer-readable instructions for implementing a method according to the present application. Thus, the present application also covers a recording medium storing a program for executing the method according to the present application, for example, a computer-readable storage medium 500 shown in fig. 5.
The methods and apparatus, electronic devices, and computer-readable storage media of the present application may be implemented in a number of ways. For example, the methods and apparatus, electronic devices, and computer-readable storage media of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order for the steps of the method is for illustration only, and the steps of the method of the present application are not limited to the order specifically described above unless specifically stated otherwise. Further, in some embodiments, the present application may also be embodied as a program recorded in a recording medium, the program including computer-readable instructions for implementing a method according to the present application. Thus, the present application also covers a recording medium storing a program for executing the method according to the present application.
The description of the present application has been presented for purposes of illustration and description, and is not intended to be exhaustive or limited to the application in the form disclosed. Many modifications and variations will be apparent to practitioners skilled in this art. The embodiment was chosen and described in order to best explain the principles of the application and the practical application, and to enable others of ordinary skill in the art to understand the application for various embodiments with various modifications as are suited to the particular use contemplated.

Claims (24)

1. A method for determining a bounding box for a target object, comprising:
acquiring attribute information of each key point in a plurality of key points of a target object in an image; the key points comprise key points of a plurality of human body parts representing human body posture forms; the attribute information of the key point includes: coordinate information and a presence discrimination value representing whether the key point is visible in the image;
determining the position of an outer frame of the target object according to the attribute information of each key point in the plurality of key points of the target object and a preset neural network, wherein the determining comprises the following steps:
determining at least one effective key point from the plurality of key points according to the attribute information of each key point in the plurality of key points;
processing the attribute information of the plurality of key points according to the attribute information of each effective key point in the at least one effective key point to obtain the processed attribute information of the plurality of key points;
inputting the processed attribute information of the plurality of key points into the preset neural network for processing to obtain the position of an outer frame of the target object;
wherein the processed attribute information of the plurality of key points includes: the processed attribute information of each effective key point in the at least one effective key point and the attribute information of other key points except the at least one effective key point in the plurality of key points.
2. The method of claim 1, wherein the target object comprises: a human body.
3. The method according to claim 1, wherein the processing the attribute information of the plurality of key points according to the attribute information of each of the at least one valid key point to obtain the processed attribute information of the plurality of key points comprises:
determining a reference coordinate according to coordinate information included in the attribute information of each effective key point in the at least one effective key point;
and determining coordinate information in the processed attribute information of each effective key point according to the reference coordinate and the coordinate information in the attribute information of each effective key point in the at least one effective key point.
4. The method according to claim 3, wherein determining the reference coordinates according to the coordinate information included in the attribute information of each of the at least one valid key point comprises:
carrying out average processing on coordinates corresponding to the coordinate information of each effective key point in the at least one effective key point to obtain the reference coordinates; and/or
Determining coordinate information in the processed attribute information of each effective key point according to the reference coordinate and coordinate information in the attribute information of each effective key point in the at least one effective key point, including:
and determining the processed coordinate information corresponding to the coordinate information of each effective key point in the at least one effective key point by taking the reference coordinate as an origin.
5. The method according to claim 3, wherein the inputting the processed attribute information of the plurality of key points into the preset neural network for processing to obtain the outer frame position of the target object includes:
inputting the processed attribute information of the plurality of key points into the preset neural network for processing to obtain output position information;
and determining the position of an outer frame of the target object according to the reference coordinate and the output position information.
6. The method according to any one of claims 1 to 5, further comprising:
obtaining a sample set comprising a plurality of sample data, wherein the sample data comprises: attribute information of a plurality of key points of a sample object, wherein the sample data is marked with an outer frame position of the sample object;
and training the neural network according to the attribute information of a plurality of key points of the sample object in each sample data and the position of the external frame of the sample object.
7. The method of any one of claims 1 to 5, wherein the neural network is trained based on a stochastic gradient descent algorithm.
8. The method of any of claims 1 to 5, wherein the target object's bounding box location comprises: and coordinate information of two vertexes in the diagonal direction of the circumscribed frame of the target object.
9. The method of any one of claims 1 to 5, wherein the neural network comprises: at least two fully-connected layers.
10. The method of any one of claims 1 to 5, wherein the neural network comprises: a three-layer fully-connected layer, wherein an activation function of at least one of a first layer fully-connected layer and a second layer fully-connected layer of the three-layer fully-connected layer comprises: the linear element ReLu activation function is modified.
11. The method of claim 10, wherein the first fully-connected layer comprises 320 neurons, the second fully-connected layer comprises 320 neurons, and a last fully-connected layer of the three fully-connected layers comprises 4 neurons.
12. An apparatus for determining a bounding box for a target object, comprising:
the acquisition module is used for acquiring attribute information of each key point in a plurality of key points of the target object in the image; the key points comprise key points of a plurality of human body parts representing human body posture forms; the attribute information of the key point includes: coordinate information and a presence discrimination value representing whether the key point is visible in the image;
the determining module is used for determining the position of an outer frame of the target object according to the attribute information of each key point in the plurality of key points of the target object acquired by the acquiring module and a preset neural network;
the determining module comprises:
the first sub-module is used for determining at least one effective key point from the plurality of key points according to the attribute information of each key point in the plurality of key points acquired by the acquisition module;
the second submodule is used for processing the attribute information of the plurality of key points according to the attribute information of each effective key point in at least one effective key point determined by the first submodule to obtain the processed attribute information of the plurality of key points;
the third sub-module is used for inputting the processed attribute information of the plurality of key points obtained by the second sub-module into the preset neural network for processing to obtain the position of an outer frame of the target object;
wherein the processed attribute information of the plurality of key points includes: the processed attribute information of each effective key point in the at least one effective key point and the attribute information of other key points except the at least one effective key point in the plurality of key points.
13. The apparatus of claim 12, wherein the target object comprises: a human body.
14. The apparatus of claim 12, wherein the second sub-module comprises:
a first unit, configured to determine a reference coordinate according to coordinate information included in attribute information of each effective key point in at least one effective key point determined by the first sub-module;
and the second unit is used for determining the coordinate information in the processed attribute information of each effective key point according to the reference coordinate determined by the first unit and the coordinate information in the attribute information of each effective key point in the at least one effective key point.
15. The apparatus according to claim 14, wherein the first unit is specifically configured to:
averaging the coordinates corresponding to the coordinate information of each effective key point in the at least one effective key point determined by the first submodule to obtain the reference coordinates; and/or
The second unit is specifically configured to:
and determining the processed coordinate information corresponding to the coordinate information of each effective key point in the at least one effective key point by using the reference coordinate determined by the first unit as an origin.
16. The apparatus of claim 14, wherein the third sub-module is specifically configured to:
inputting the processed attribute information of the plurality of key points obtained by the second submodule into the preset neural network for processing to obtain output position information;
and determining the position of an outer frame of the target object according to the reference coordinate and the output position information.
17. The apparatus of any one of claims 12 to 16, further comprising: a training module to:
obtaining a sample set comprising a plurality of sample data, wherein the sample data comprises: attribute information of a plurality of key points of a sample object, wherein the sample data is marked with an outer frame position of the sample object;
and training the neural network according to the attribute information of a plurality of key points of the sample object in each sample data and the position of the external frame of the sample object.
18. The apparatus of any one of claims 12 to 16, wherein the neural network is trained based on a stochastic gradient descent algorithm.
19. The apparatus of any of claims 12 to 16, wherein the target object's outer frame position comprises: and coordinate information of two vertexes in the diagonal direction of the circumscribed frame of the target object.
20. The apparatus of any one of claims 12 to 16, wherein the neural network comprises: at least two fully-connected layers.
21. The apparatus of any one of claims 12 to 16, wherein the neural network comprises: a three-layer fully-connected layer, wherein an activation function of at least one of a first layer fully-connected layer and a second layer fully-connected layer of the three-layer fully-connected layer comprises: the linear element ReLu activation function is modified.
22. The apparatus of claim 21, wherein the first fully-connected layer comprises 320 neurons, the second fully-connected layer comprises 320 neurons, and a last fully-connected layer of the three fully-connected layers comprises 4 neurons.
23. An electronic device, comprising: a processor and a computer-readable storage medium to store instructions, execution of which by the processor causes the electronic device to perform the method of any of claims 1-11.
24. A computer readable storage medium having stored thereon instructions which, when executed by a processor, perform the method of any of claims 1 to 11.
CN201711165979.8A 2017-11-21 2017-11-21 Method and device for determining outer frame of target object and electronic equipment Active CN108229305B (en)

Priority Applications (5)

Application Number Priority Date Filing Date Title
CN201711165979.8A CN108229305B (en) 2017-11-21 2017-11-21 Method and device for determining outer frame of target object and electronic equipment
SG11201913529UA SG11201913529UA (en) 2017-11-21 2018-10-23 Methods and apparatuses for determining bounding box of target object, media, and devices
PCT/CN2018/111464 WO2019100886A1 (en) 2017-11-21 2018-10-23 Method and apparatus for determining external frame of target object, medium, and device
JP2019572712A JP6872044B2 (en) 2017-11-21 2018-10-23 Methods, devices, media and equipment for determining the circumscribed frame of an object
US16/731,858 US11348275B2 (en) 2017-11-21 2019-12-31 Methods and apparatuses for determining bounding box of target object, media, and devices

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711165979.8A CN108229305B (en) 2017-11-21 2017-11-21 Method and device for determining outer frame of target object and electronic equipment

Publications (2)

Publication Number Publication Date
CN108229305A CN108229305A (en) 2018-06-29
CN108229305B true CN108229305B (en) 2021-06-04

Family

ID=62652771

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711165979.8A Active CN108229305B (en) 2017-11-21 2017-11-21 Method and device for determining outer frame of target object and electronic equipment

Country Status (5)

Country Link
US (1) US11348275B2 (en)
JP (1) JP6872044B2 (en)
CN (1) CN108229305B (en)
SG (1) SG11201913529UA (en)
WO (1) WO2019100886A1 (en)

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2018033137A1 (en) * 2016-08-19 2018-02-22 北京市商汤科技开发有限公司 Method, apparatus, and electronic device for displaying service object in video image
CN108229305B (en) * 2017-11-21 2021-06-04 北京市商汤科技开发有限公司 Method and device for determining outer frame of target object and electronic equipment
CN110826357B (en) * 2018-08-07 2022-07-26 北京市商汤科技开发有限公司 Method, device, medium and equipment for three-dimensional detection and intelligent driving control of object
CN111241887B (en) * 2018-11-29 2024-04-16 北京市商汤科技开发有限公司 Target object key point identification method and device, electronic equipment and storage medium
CN112115746B (en) * 2019-06-21 2024-06-18 富士通株式会社 Human body action recognition device and method and electronic equipment
CN110782404B (en) * 2019-10-11 2022-06-10 北京达佳互联信息技术有限公司 Image processing method, device and storage medium
CN110929792B (en) * 2019-11-27 2024-05-24 深圳市商汤科技有限公司 Image labeling method, device, electronic equipment and storage medium
CN113780040B (en) * 2020-06-19 2024-10-22 北京沃东天骏信息技术有限公司 Positioning method and device for lip key points, storage medium and electronic equipment

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107194361A (en) * 2017-05-27 2017-09-22 成都通甲优博科技有限责任公司 Two-dimentional pose detection method and device
CN107220604A (en) * 2017-05-18 2017-09-29 清华大学深圳研究生院 A kind of fall detection method based on video

Family Cites Families (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP5665401B2 (en) * 2010-07-21 2015-02-04 キヤノン株式会社 Image processing apparatus, image processing method, and program
US9342888B2 (en) * 2014-02-08 2016-05-17 Honda Motor Co., Ltd. System and method for mapping, localization and pose correction of a vehicle based on images
IL231862A (en) * 2014-04-01 2015-04-30 Superfish Ltd Neural network image representation
JP2016006626A (en) 2014-05-28 2016-01-14 株式会社デンソーアイティーラボラトリ Detector, detection program, detection method, vehicle, parameter calculation device, parameter calculation program, and parameter calculation method
WO2016004330A1 (en) * 2014-07-03 2016-01-07 Oim Squared Inc. Interactive content generation
CN104573715B (en) 2014-12-30 2017-07-25 百度在线网络技术（北京）有限公司 The recognition methods in image subject region and device
CN107851192B (en) 2015-05-13 2023-04-14 北京市商汤科技开发有限公司 Apparatus and method for detecting face part and face
US9767381B2 (en) 2015-09-22 2017-09-19 Xerox Corporation Similarity-based detection of prominent objects using deep CNN pooling layers as features
WO2017095948A1 (en) 2015-11-30 2017-06-08 Pilot Ai Labs, Inc. Improved general object detection using neural networks
KR102592076B1 (en) 2015-12-14 2023-10-19 삼성전자주식회사 Appartus and method for Object detection based on Deep leaning, apparatus for Learning thereof
CN107194338A (en) 2017-05-14 2017-09-22 北京工业大学 Traffic environment pedestrian detection method based on human body tree graph model
US11080551B2 (en) * 2017-05-22 2021-08-03 Intel Corporation Proposal region filter for digital image processing
US10438371B2 (en) * 2017-09-22 2019-10-08 Zoox, Inc. Three-dimensional bounding box from two-dimensional image and point cloud data
CN108229305B (en) * 2017-11-21 2021-06-04 北京市商汤科技开发有限公司 Method and device for determining outer frame of target object and electronic equipment

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107220604A (en) * 2017-05-18 2017-09-29 清华大学深圳研究生院 A kind of fall detection method based on video
CN107194361A (en) * 2017-05-27 2017-09-22 成都通甲优博科技有限责任公司 Two-dimentional pose detection method and device

Also Published As

Publication number Publication date
US11348275B2 (en) 2022-05-31
WO2019100886A1 (en) 2019-05-31
CN108229305A (en) 2018-06-29
JP2020525959A (en) 2020-08-27
JP6872044B2 (en) 2021-05-19
SG11201913529UA (en) 2020-01-30
US20200134859A1 (en) 2020-04-30

Similar Documents

Publication Publication Date Title
CN108229305B (en) Method and device for determining outer frame of target object and electronic equipment
CN108427927B (en) Object re-recognition method and apparatus, electronic device, program, and storage medium
US11120254B2 (en) Methods and apparatuses for determining hand three-dimensional data
CN108229343B (en) Target object key point detection method, deep learning neural network and device
US10977523B2 (en) Methods and apparatuses for identifying object category, and electronic devices
CN107358149B (en) Human body posture detection method and device
CN108229301B (en) Eyelid line detection method and device and electronic equipment
CN109711508B (en) Image processing method and device
CN108304775A (en) Remote sensing images recognition methods, device, storage medium and electronic equipment
CN108573471B (en) Image processing apparatus, image processing method, and recording medium
CN110222641B (en) Method and apparatus for recognizing image
CN111459269B (en) Augmented reality display method, system and computer readable storage medium
US20190197727A1 (en) Object recognition processing apparatus, object recognition processing method, and program
CN112927353A (en) Three-dimensional scene reconstruction method based on two-dimensional target detection and model alignment, storage medium and terminal
CN108229494B (en) Network training method, processing method, device, storage medium and electronic equipment
JP6658188B2 (en) Image processing apparatus, image processing method, and image processing program
CN110832542A (en) Recognition processing device, recognition processing method, and program
WO2023083030A1 (en) Posture recognition method and related device
CN109345460B (en) Method and apparatus for rectifying image
EP3410389A1 (en) Image processing method and device
CN109934058B (en) Face image processing method, face image processing device, electronic apparatus, storage medium, and program
CN110414458B (en) Positioning method and device based on matching of plane label and template
CN113793349A (en) Target detection method and device, computer readable storage medium and electronic equipment
CN114694263B (en) Action recognition method, device, equipment and storage medium
CN111968030B (en) Information generation method, apparatus, electronic device and computer readable medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant